The Coherence window is the RR-derived coherence operator surface. It uses the
accepted RR interval stream, builds a rolling resonance window, and surfaces a
smoothed normalized coherence value with confidence and underlying band-power telemetry.
Reference Model
The fixed spectral method follows McCraty, Atkinson, Tomasino, and Bradley, The Coherent Heart: Heart-Brain Interactions, Psychophysiological Coherence, and the Emergence of System-Wide Order (Institute of HeartMath, 2006).
The app keeps these paper-defined constants fixed:
- search the maximum spectral peak in
0.04-0.26 Hz - integrate peak power in a
0.030 Hzwindow centered on that peak - integrate total power across
0.0033-0.4 Hz - compute the paper ratio as
(Peak Power / (Total Power - Peak Power))^2
The app also keeps an operator-facing normalized score for the headline readout and summary charts:
- normalized score =
Peak Power / Total Power, clamped to0..1
The paper defines the spectral calculation above, but it does not define the operator-side runtime controls exposed by this app.
Real-vs-synthetic alignment note:
- on a physical H10, the RR intervals feeding coherence come from the strap's device-side beat timing, which is derived from ECG
- on the sibling
SyntheticBiotransport, the coherence tracker still solves from RR, but the transport now also emits synthetic PMD ECG frames that stay synchronized with that RR schedule so the waveform view does not drift away from the interval view during synthetic sessions
Open formula sheet Download Markdown Download PDF
What The Window Exposes
- headline
Coherencevalue plusConfidence - tracker readiness, stale state, stabilization progress, and RR sample count
- current heartbeat BPM and the last accepted RR interval
- peak frequency, peak band power, total band power, and raw coherence
- reset, apply-tuning, and restore-defaults controls for the current device
- a dedicated chart and log stream for the selected device
Manual App Test
- Wear the strap and wet the electrodes.
- Start the app and connect to the intended strap.
- Confirm HR updates and RR intervals are moving on the
Telemetrytab. - Open
Coherence windowfrom theTelemetrytab. - Check the top status line:
- before the RR window stabilizes, coherence can stay unavailable
- confidence should remain low until enough accepted RR data accumulates
- Keep the strap still enough for clean RR timing while the warmup runs.
- Confirm the coherence window starts updating:
- the coherence headline leaves
-- - confidence rises above zero
- RR sample count and coverage increase
- peak frequency and band-power fields populate once a coherence sample exists
- the coherence headline leaves
- Use the shared telemetry selector if you want the
Telemetrytab to plot:CoherenceCoherence confidence
- Adjust tuning only if needed:
Apply tuningresets the tracker with the new RR window settingsRestore defaultsreturns to the app defaultsReset trackerclears the current RR-derived window state
Tuning Parameters
Min RR samples: accepted RR intervals required before the tracker attempts a spectral solveWindow (s): rolling RR history length used for spline resampling and PSD analysisSmoothing speed: exponential smoothing speed applied to the displayed normalized scoreStale timeout (s): freshness timeout for theTrackingstate once RR updates stop
The following coherence constants are intentionally not user-tunable in the app:
- peak-search band
0.04-0.26 Hz - peak-window width
0.030 Hz - total-band range
0.0033-0.4 Hz
Fast Failure Checks
- If the window stays on
No RRorAwaiting device selection, HR may be present but usable RR intervals are not reaching the tracker yet. - If the status says it is still building the RR coherence window, keep the stream running longer before treating it as a failure.
- If coherence becomes available and then goes stale, the RR input is no longer fresh enough for the configured timeout.
- If confidence stays near zero, the tracker has not collected enough stable RR coverage yet.